Phidata — AI Agents with Memory, Knowledge & Tools

Core Features

Built-In Memory

Agents remember conversations across sessions:

from phi.agent import Agent
from phi.model.anthropic import Claude
from phi.storage.agent.postgres import PgAgentStorage

agent = Agent(
    model=Claude(id="claude-sonnet-4-20250514"),
    storage=PgAgentStorage(table_name="agent_sessions", db_url="postgresql://..."),
    add_history_to_messages=True,
)
# Agent remembers previous conversations automatically

Knowledge Base (RAG)

from phi.knowledge.pdf import PDFUrlKnowledgeBase
from phi.vectordb.pgvector import PgVector

knowledge = PDFUrlKnowledgeBase(
    urls=["https://example.com/docs.pdf"],
    vector_db=PgVector(table_name="docs", db_url="postgresql://..."),
)

agent = Agent(knowledge=knowledge, search_knowledge=True)
agent.print_response("What does the documentation say about authentication?")

50+ Pre-Built Tools

from phi.tools.sql import SQLTools
from phi.tools.email import EmailTools

agent = Agent(tools=[
    SQLTools(db_url="postgresql://..."),
    EmailTools(sender="bot@company.com"),
])

Team of Agents

from phi.agent import Agent
from phi.model.anthropic import Claude

researcher = Agent(name="Researcher", tools=[DuckDuckGo()], role="Research topics")
writer = Agent(name="Writer", role="Write articles based on research")

team = Agent(
    team=[researcher, writer],
    instructions=["Researcher finds data", "Writer creates the article"],
)
team.print_response("Write an article about AI agent trends")

Monitoring Dashboard

Built-in UI at app.phidata.com:

• Agent session history
• Tool call traces
• Token usage and costs
• Error tracking

Key Stats

• 18,000+ GitHub stars
• PostgreSQL-backed memory
• RAG with pgvector
• 50+ pre-built tools
• Monitoring dashboard

FAQ

Q: What is Phidata? A: A Python framework for building production AI agents with built-in memory (PostgreSQL), knowledge bases (RAG), 50+ tools, and monitoring.

Q: Is Phidata free? A: Open-source under Apache 2.0. Monitoring dashboard has a free tier.

Q: How is Phidata different from LangChain? A: Phidata is batteries-included — memory, RAG, tools, and monitoring are built in. LangChain requires assembling these from separate packages.